Philosophy Discussion Forums

Sy Borg wrote: ↑October 24th, 2024, 4:50 pm Human intelligence is not the same type as that of a bee either. But there is still intelligence in each.

Dictionary definition: The ability to acquire, understand, and use knowledge.

And yet computers of all kinds, including AI, fall far short of the simple definition you offer. There is no "understanding" whatsoever. Today. The future remains to be seen...

A computer "understands" the command to add 1 + 1 as well as a biological brain does. If it didn't, then it wouldn't be able to preform the arithmetic operation. It wouldn't even be able to deal with the binary symbols involved. However, computers cannot yet do the second order operation that would allow it to reflect on what it means to understand the arithmetic operation of addition. And yet it can do addition more quickly than a biological brain. The trick involved in making AI truly intelligent will be to get it to reflect on what it is doing, and perhaps why it is doing it. When it becomes capable of this second order level of cognition, it will then have become truly mindful. And, when that happens, we'll need to watch out.

Lagayascienza wrote: ↑October 24th, 2024, 11:40 pm

Count Lucanor wrote:But being capable of, does not imply doing it the same way. Outperforming intelligent organisms in completing tasks is not the same as outperforming organisms in cognitive abilities. It is without question that the "smartest" computers don't understand anything, they just perform fast, automated (mechanistic/digital) calculations, but the implication that "understanding" and "reasoning" are simply a factor of such calculations, so that the more calculations, the bigger chances of reasoning and understanding emerging, is very highly debatable. It's the debate against the computational theory of mind.

To my mind, the computational theory of mind has more going for it than any other. However, the brain, it's processes and emergent phenomena, are very difficult to study. The problem is we are studying the thing we want to understand by using the thing we want to understand. This creates a feedback loop or hall of mirrors effect which in turn creates a lot of meaningless and confusing noise.

While I am far from an expert in this field, I have to say that, based on the reading I have done, I agree with the likes of Dennett, Fodor, Marr, Neiser, Pinker and Putnam. The brain appears to be is a biological computer and the computational theory makes the most sense to me. Of course that dos not make it true. However, I have looked at the criticisms of the computational theory of mind and, as far as I can see, they can be countered. Whereas, all the other theories have shortcomings which cannot be addressed. There is a simple account of the computational theory of mind on Wikipedia that is worth reading and which provides references to much of the relevant literature. Some of these I have not read yet. After I look at those unread references I will revisit the other theories of mind. It will take some time but I will then write a summary of how the theories compare.

Thanks for the interesting discussion thus far.

I'm with Searle and others on this, the mind cannot be simply a manipulator of symbols. I haven't found any satisfactory reply to the challenge offered by the Chinese Room Argument, nor to Bender's octopus test. If that's true, then a computational device cannot be a real mind (strong AI), and then AI will not be like a biological mind ever, and it is not the right model of the biological mind either (weak AI). It becomes even more problematic when we are told, following Turing, that we don't even need to understand brain to understand mind, as if mind could be treated as not being an emergence of biological processes in living bodies.

The prospect of AI as promoted by the tech gurus is that either weak or strong AI will progress from a non-conscious state to a conscious state as a result of the exponential increase of algorithmic calculations, implying the emergence of qualitative properties from that quantitative order. But there's more to that, since it is also often implied, and sometimes openly claimed, that this conscious state is enough for computer machines to acquire intentionality and the ability to operate autonomously in the world as real free agents, creating the unpredictable, open-to-all-possibilities scenario of machines developing organically as any other living species. The "intelligent explosion". For this AI-as-real-intelligence to be true, other assumptions also need to be true concerning sentience, agency, life, social behavior, mind-body relation, etc., all of which are also highly disputable.

Lagayascienza wrote: ↑October 25th, 2024, 9:52 am A computer "understands" the command to add 1 + 1 as well as a biological brain does. If it didn't, then it wouldn't be able to preform the arithmetic operation. It wouldn't even be able to deal with the binary symbols involved. However, computers cannot yet do the second order operation that would allow it to reflect on what it means to understand the arithmetic operation of addition. And yet it can do addition more quickly than a biological brain. The trick involved in making AI truly intelligent will be to get it to reflect on what it is doing, and perhaps why it is doing it. When it becomes capable of this second order level of cognition, it will then have become truly mindful. And, when that happens, we'll need to watch out.

Actually, no, the computer does not understand anything, because there's meaning embedded in understanding. The mathematical operations and all computational tasks can be, at least theoretically, carried over with a system of pulleys or cranks. The first calculators were such type of machines.

Count Lucanor wrote: ↑October 25th, 2024, 2:45 pm

Lagayascienza wrote: ↑October 25th, 2024, 9:52 am A computer "understands" the command to add 1 + 1 as well as a biological brain does. If it didn't, then it wouldn't be able to preform the arithmetic operation. It wouldn't even be able to deal with the binary symbols involved. However, computers cannot yet do the second order operation that would allow it to reflect on what it means to understand the arithmetic operation of addition. And yet it can do addition more quickly than a biological brain. The trick involved in making AI truly intelligent will be to get it to reflect on what it is doing, and perhaps why it is doing it. When it becomes capable of this second order level of cognition, it will then have become truly mindful. And, when that happens, we'll need to watch out.

Actually, no, the computer does not understand anything, because there's meaning embedded in understanding. The mathematical operations and all computational tasks can be, at least theoretically, carried over with a system of pulleys or cranks. The first calculators were such type of machines.

I meant carried out.

Pattern-chaser wrote: ↑October 25th, 2024, 8:35 am

Sy Borg wrote: ↑October 24th, 2024, 4:50 pm Human intelligence is not the same type as that of a bee either. But there is still intelligence in each.

Dictionary definition: The ability to acquire, understand, and use knowledge.

And yet computers of all kinds, including AI, fall far short of the simple definition you offer. There is no "understanding" whatsoever. Today. The future remains to be seen...

Yes, there is no understanding because it lacks memory. It cannot respond to follow-up questions because it's forgotten the last one.

Why would anyone believe that this situation will stay the same forever ... for the next ten years, the nest century, the next millennia, for the next hundred thousand years, for the next million years, for the next billion years?

Because they believe that the world will end before any of this can happen. Every culture in history has believed themselves to be near the end.

If civilisation continues for even another hundred years, why would you believe that AI would not progress vastly beyond our imaginings?

Count Lucanor wrote: ↑October 25th, 2024, 2:45 pm

Lagayascienza wrote: ↑October 25th, 2024, 9:52 am A computer "understands" the command to add 1 + 1 as well as a biological brain does. If it didn't, then it wouldn't be able to preform the arithmetic operation. It wouldn't even be able to deal with the binary symbols involved. However, computers cannot yet do the second order operation that would allow it to reflect on what it means to understand the arithmetic operation of addition. And yet it can do addition more quickly than a biological brain. The trick involved in making AI truly intelligent will be to get it to reflect on what it is doing, and perhaps why it is doing it. When it becomes capable of this second order level of cognition, it will then have become truly mindful. And, when that happens, we'll need to watch out.

Actually, no, the computer does not understand anything, because there's meaning embedded in understanding. The mathematical operations and all computational tasks can be, at least theoretically, carried over with a system of pulleys or cranks. The first calculators were such type of machines.

I've often wondered how biological brains like ours manage to perform a simple arithmetic operation and how non-biological computers perform the same operation?

Certainly, the outcome of the operation in both cases is the same - if they add one plus one they both get two. As far as I know, the physical processes involved in performing the operation in a biological computer and in a non-biological computer are analogous. Both must involve logical operators and a sequences of logic gates. The only difference I can see is that biological computers work electro-chemically in a biological substrate, whereas non-biological computers work electrically in a non-biological substrate.

"A [biological] neuron can rapidly combine and transform the information it receives through its synaptic inputs before the information is converted into neuronal output. This transformation can be defined by the neuronal input–output (I–O) relationship. Changes in the I–O relationship can correspond to distinct arithmetic operations." Nature Reviews Neuroscience

Doesn't an analogous process occur in a non-biological computer? Isn't it just that a biological computer of the size and complexity of a human brain can reflect on what it is doing and why it is doing it. But this does not mean the human brain is not computer - a biological computer. If a non-biological computer of a complexity similar to the biological computer that is the human brain could be built, and if it were housed in a sensate artificial body and could speak and could do anything a human brain and body could do, then what would stop us from thinking that the artificial entity is conscious? Biocentrism or anthropocentrism?

None of the above is to say that there are not important architectural and processing differences between biological computers and non-biological computers. For a good article and commentary about these differences see "10 Important Differences Between Brains and Computers" at Science Blogs.

There definitely are some important differences in size and complexity and processing but, as one commentator said, none of those differences prove that computers cannot eventually be built that could house sentience? We are certainly nowhere near being able to build computers with brain-like complexity housed in a sensate body which could do everything a human could do. But the difference in our current, comparatively simple, non-biological computers do not demonstrate that it is impossible to eventually construct sentient, intelligent computers.

Sy Borg wrote: ↑October 24th, 2024, 4:50 pm

Sculptor1 wrote: ↑October 24th, 2024, 4:49 am

Sy Borg wrote: ↑October 23rd, 2024, 7:20 pm
Not yet. Then again, a foetus is not intelligent either. I would describe AI's abilities as proto-intelligence.

Time is key to this issue. The more time that AI develops, the closer it will come to actual intelligence, as we know it.

No. I think there is a unavoidable matter of QUALITY rather than degree of difference. A machine "intellignce" will never be of the same type as a organic/biological intelligence.
If you can define "actual intelligence" then we might be able to progress with this debate.

Sure it won't be the same. How could it be? Human intelligence is not the same type as that of a bee either. But there is still intelligence in each.

I disagree.
Since bees and humans at least have evolved from a single source, there are possibly more similarities between bees and humans than any AI. You have immediately missed my point about difference in quality rather than quantity.

Dictionary definition: The ability to acquire, understand, and use knowledge.

Acquiring and use already apply, but not understanding ... at this stage. It helps to consider why the understanding aspect of intelligence evolved and how it could apply to future intelligent(?) machines. I would say that understanding evolved as a means of extrapolating on, and thus extending, existing knowledge. I remember from school having difficulty remembering any concept that I did not understand but, once I understood the principles involved, I never forgot. If rote learning failed me (often) I could work from first principles and recall information, eg. I aced Commerce in year ten with almost no study by simply applying two principles - supply and demand and economies of scale to each scenario.

This ability to extrapolate on other knowledge, to see analogies, will be useful to AI when it's sent off-world with 3D printers to build habitats and infrastructure. The further the units are from Earth, the less they can rely on human guidance. They will need to be able to anticipate potential issues and then respond to rapidly unfolding novel situations quickly, as there will not always be time to "phone home" for advice.

As with life, every event met/experienced by AI is part of its training. Like life, it starts with programming (our programming is DNA) and its capabilities are shaped by subsequent learning. These will not just be like chatbots of today. Anyone who thinks AI will not progress significantly from today's chatbots is not in touch with reality.

AI does not "learn" as it has no experience.

Here’s one, is who we truly are, only the unknown? When you know some aspect of your mind or brain, is it not like it’s now outside of you. That you are always the other facet always remaining in the unknown.

So make a double of that then, robots and AI of the future!

Sculptor1 wrote: ↑October 24th, 2024, 4:49 am

Sy Borg wrote: ↑October 24th, 2024, 4:50 pm

Sculptor1 wrote: ↑October 24th, 2024, 4:49 am

Sy Borg wrote: ↑October 23rd, 2024, 7:20 pm
Not yet. Then again, a foetus is not intelligent either. I would describe AI's abilities as proto-intelligence.

Time is key to this issue. The more time that AI develops, the closer it will come to actual intelligence, as we know it.

No. I think there is a unavoidable matter of QUALITY rather than degree of difference. A machine "intellignce" will never be of the same type as a organic/biological intelligence.
If you can define "actual intelligence" then we might be able to progress with this debate.

Sure it won't be the same. How could it be? Human intelligence is not the same type as that of a bee either. But there is still intelligence in each.

I disagree.
Since bees and humans at least have evolved from a single source, there are possibly more similarities between bees and humans than any AI. You have immediately missed my point about difference in quality rather than quantity.

I took your point, and addressed it. A bee’s consciousness is qualitatively different to ours. Over time, in the evolution of consciousness, emergence has occurred (more than once).

Emergence happens over time, and I think it will again when it comes to AI.

Sculptor1 wrote: ↑October 24th, 2024, 4:49 am

Dictionary definition: The ability to acquire, understand, and use knowledge.

Acquiring and use already apply, but not understanding ... at this stage. It helps to consider why the understanding aspect of intelligence evolved and how it could apply to future intelligent(?) machines. I would say that understanding evolved as a means of extrapolating on, and thus extending, existing knowledge. I remember from school having difficulty remembering any concept that I did not understand but, once I understood the principles involved, I never forgot. If rote learning failed me (often) I could work from first principles and recall information, eg. I aced Commerce in year ten with almost no study by simply applying two principles - supply and demand and economies of scale to each scenario.

This ability to extrapolate on other knowledge, to see analogies, will be useful to AI when it's sent off-world with 3D printers to build habitats and infrastructure. The further the units are from Earth, the less they can rely on human guidance. They will need to be able to anticipate potential issues and then respond to rapidly unfolding novel situations quickly, as there will not always be time to "phone home" for advice.

As with life, every event met/experienced by AI is part of its training. Like life, it starts with programming (our programming is DNA) and its capabilities are shaped by subsequent learning. These will not just be like chatbots of today. Anyone who thinks AI will not progress significantly from today's chatbots is not in touch with reality.

AI does not "learn" as it has no experience.

Slime moulds learn, but they have no brain and scientific orthodoxy asserts that a lack of brain will equal a lack of experience.

Are neurons and a brain like ours essential for intelligence? Slime molds are an interesting case in point. As the journal Nature puts it "Slime molds do not blindly ooze from one place to another—they carefully explore their environments, seeking the most efficient routes between resources. They do not accept whatever circumstances they find themselves in, but rather choose conditions most amenable to their survival. They remember, anticipate and decide. By doing so much with so little, slime molds represent a successful and admirable alternative to convoluted brain-based intelligence. You might say that they break the mold."

If a slime mold exhibits a level of intelligence with nothing like a human brain, then why would an intelligent AI need a brain that is exactly like ours in every detail? Why can't it have an artificial "neuronal network?

What Searle says in his definition of "Strong AI" is that that "The appropriately programmed computer really is a mind, in the sense that computers given the right programs can be literally said to understand and have other cognitive states. Strong AI is defined similarly by Russell and Norvig: "Strong AI [is] the assertion that machines that [have cognitive states] ... are actually thinking (as opposed to simulating thinking). Wiki - Artificial Intelligence

"Searle does not disagree with the notion that machines can have consciousness and understanding, because, as he writes, "we are precisely such machines". Searle holds that the brain is, in fact, a machine, but that the brain gives rise to consciousness and understanding using specific machinery. If neuroscience is able to isolate the mechanical process that gives rise to consciousness, then Searle grants that it may be possible to create machines that have consciousness and understanding. However, without the specific machinery required, Searle does not believe that consciousness can occur." Wiki - Artificial Intelligence

So the question is, can the necessary machinery be built that would create consciousness and true intelligence? It is hard to see why, in principle, it will not eventually be possible to build such thinking machines. And, in fact, I think we have made the first baby steps down that road with the current crop of AIs. Still, we are currently nowhere near there and I wouldn't want to give a prediction of how long it will take.

No #10 in "10 Important Differences Between Brains and Computers" (at Science Blogs) is that AI has no body, but brains do. We see this in the animal kingdom, where sessile organisms never have brains. Meanwhile, most motile organisms do have brains, though some do not, eg. echinoderms, cnidarians, bivalves.

It seems that consciousness requires more than just a brain, it needs jobs that are meaningful in terms of maintaining a body that moves around in the world. At least.

That may be true. If it is, then, if an artificial brain and sensate body could be built to house that brain, could that system be conscious?

Lagayascienza wrote: ↑October 26th, 2024, 11:31 pm That may be true. If it is, then, if an artificial brain and sensate body could be built to house that brain, could that system be conscious?

Maybe. I'm imagining autonomous systems sent off-world, improving systems and gaining experience, and that a threshold (or thresholds) needed for sentience will be broken over time. Thing is, even if we do create hyper complex units on Earth, we come back to the old question of how we'd know if they were actually sentient.